Limits stop a request before it starts. Nothing is charged for a request that did not run, and Keplar does not swap in a cheaper model to hide that you are at a limit.
Which limit did I hit?
| Message says | Limit | What to do |
|---|---|---|
| Daily question limit | Free plan's 50-a-day allowance | Wait for the reset, or upgrade to Lite, then Plus |
| 5-hour limit, with a countdown | Rolling short window on a paid plan | Wait, use Fastest, or upgrade |
| Weekly limit, with a reset day | Rolling weekly window | Wait for the reset day or upgrade |
| Not enough usage for Deep Research | Monthly allowance cannot cover a run | Ask without Deep Research, wait for reset, or upgrade |
| Too many requests | Per-minute rate limit | Wait a minute and try again |
| The free models are busy | Upstream capacity, not your allowance | Try again shortly, or upgrade |
What still works
Small talk is free everywhere. Reading old chats, editing sites you already built and downloading exports do not use credits.
Keep going cheaply
- Choose Fastest for low-stakes questions.
- Start a new chat for a new topic instead of carrying a long history.
- Turn off Deep Research unless you need it.
- Ask one clear question instead of several vague ones.
Reset times
Monthly credits reset at your billing-period end. The 5-hour and weekly windows reset on the schedule shown next to the meter. The Free daily count resets daily.
If you think a limit is wrong
Failed runs and runs with no answer are never counted. If your meter looks off, email team@keplar.one with the time and what you asked (not the content, if sensitive).
Related
- Rolling usage limits: The 5-hour and weekly limits on paid plans: how a window opens and resets, the numbers per plan, the warning at 80 percent, and what to do when you hit one.
- Credits explained: What a Keplar credit is, what costs more or less, how credits are counted from real provider charges, and what does not use credits.
- Messages and what to do: The messages you may see in Keplar, such as busy free models, too-large questions, attachments, no vision models and generic failures, with causes and fixes.